Thread by @neutronsNeurons

Something interesting happened last week to one of our students, @perryrjohnson.

Gather round folks. Thread coming up. 👇

1/ Perry likes to hike in the wilderness. And he likes to plan his hikes with an app.

Let's call it HikingApp.

For reasons that will become clear, this is not the app's real name.

2/ Perry loves HikingApp, both the product and the company.

He uses the app almost every week, and the company's vision resonates with him.

3/ Perry's a determined guy, and he's in the market for a data science job.

So Perry decides to convince HikingApp to hire him as a data scientist.

(Companies: power users of your product make *great* employees.)

4/ To get hired, Perry would need to rise above the noise of HikingApp's other applicants.

His first goal was to get their attention.
His second goal was to showcase his unique engineering skillset.

He succeeded at both. In *spectacular* fashion.

5/ Before I say more, there's something you should understand about Perry: he's a self-taught hacker.

So there are certain conventions that you or I may be aware of - around scraping, robots.txt, etc. - that Perry was *not* aware of.

This will become important shortly.

6/ Back to the story:

Perry had decided on a way to get HikingApp's attention: he would add a new feature to their product.

7/ This new feature would be a personalized recommender system for HikingApp. It would let users discover new hiking trails they hadn't explored before. Cool idea, right?

As a power user, this was a feature Perry felt HikingApp was missing.

8/ But there was a problem: Perry needed training data, and HikingApp didn't have an API.

So Perry did what an ambitious, driven, well-meaning, but totally self-taught hacker would do.

He scraped 11,000 records from HikingApp, and used those to train his recommender.

9/ Then he posted his recommender, trained on a scraped HikingApp dataset, to a public URL.

10/ *Then* he wrote a blog post describing how much he loved HikingApp,

how amazing it would be if they had a trail recommender,

and patiently explaining how to scrape their database efficiently at scale.

11/ *And then* - he emailed HikingApp's CEO, saying he was interested in working for him, and politely asking for a meeting.

Perry helpfully included links to his scraped app and blog post.

Hoping these would get the CEO's attention.

12/ Again - and I can't emphasize this enough - Perry was *not* aware of the conventions around web scraping.

He is *very acutely* aware of them now.

13/ Back to our story:

Perry's email did, in fact, succeed in getting the CEO's attention.

His complete - undivided - attention. 😬

14/ You can probably write the rest of this story yourself.

15/

9:15 am: Perry sends enthusiastic email to HikingApp CEO.

10:15 am: Perry receives a firm personal request from HikingApp executive: please take down your web app and blog post.

10:20 am: Perry facepalms; pulls his work off the Internets.

16/ HikingApp was pretty nice about it.

But they were freaked out that a self-taught hacker pulled thousands of records from their system in a few days without them noticing, even if his intentions were 100% good.

17/ Perry feels bad about it.

But if anything falls under "entire class of mistake he won't make again", this sure does.

18/ I'm sad that Perry's hard work won't see the light of day.

But there's no denying what he accomplished, or the kind of guy he is.

Perry is a guy so determined, he'll add a new feature to your product *as a way of introducing himself.*

19/ At the end of the day, anybody can learn web scraping conventions. But determination like Perry's - that can't be taught.

20/ We've got a lot of incredible students. But Perry is one in a million. I wanted to write this story so people understood why.

If you're interested in talking to Perry, his DMs are open: @perryrjohnson

He's based in Seattle. And he is one hell of a data scientist.